KINGDOM OF SAUDI ARABIA | JAZAN UNIVERSITY COLLEGE OF COMPUTER SCIENCE \& INFORMATION SYSTEMS ASSIGNMENT PROBLEM STATEMENT Guidelines: - Assignment should submit on BLACKBOARD only - Questions are from chapter 1 and 2 , You can refer the resourcos from the internet. - The last date of submission will be on - Late subnission will result in reduction of marks. Questions: 1. Construct the following. (CLO 2.1) - Single-layered feed forward network with 2 input neurons and 4 output neurons. - Multi-layered feed forward network with 3 input neurons, 6 hidden neuron and 2 outputs. - A fully recurrent network with 4 neurons, but with no self-feedback. 2. A neuron ' k ' receives input from other five neurons, as 2.65,3.85,4.2 and 0.1 The corresponding synaptic weights are - 1.5, 0.2,1.3 and 0.5. Assume the bias as -1. (CLO 1.1) - Draw the computational model of neuron ' k '. - Calculate the linear output of the neuron. - Apply sigmoid function as activation function 3. What are the different types of activation function describe them with the help of an example and diagram. (CLO 1.1) 4. How to assign credit assignment problem with two sub problems for a neural network's 5. Discuss this relationship between supervised learning and reinforcement learning.(CLO 2.2) 6. Explain in detail the method of steepest descent in single layer perceptron. (CLO 2.1) 7. In a perceptron network there are two inputs and two synaptic weights. (CLO 2.1) Classify the perceptron is Class 1 or Class 2 using decision boundary. Find the given inputs (3,7) lies above line or below line in the quadratic equation 4x+ 2y15